Validation

In order to validate our results, this notebook will show the flood mapping for the same area and time range from the ‘Global Flood Monitoring’(GFM) to allow comparisons.

import hvplot.xarray
import pystac_client
from odc import stac as odc_stac
from dask.distributed import Client, wait

client = Client(processes=False, threads_per_worker=2, n_workers=3, memory_limit="12GB")

# Connect to STAC catalog
eodc_catalog = pystac_client.Client.open("https://stac.eodc.eu/api/v1/")

# Search for available images
time_range = "2022-10-11/2022-10-25"
bounding_box = [12.3, 54.3, 13.1, 54.6]
search = eodc_catalog.search(collections="GFM", bbox=bounding_box, datetime=time_range)
items_GFM = search.item_collection()

print(f"Found {len(items_GFM)} items") 

Found 59 items

Load the data

import pyproj
crs = pyproj.CRS.from_wkt(items_GFM[0].properties["proj:wkt2"])

# Set the resolution of the data
resolution = items_GFM[0].properties['gsd']

GFM_dc= odc_stac.load(
    items_GFM, 
    bbox=bounding_box,   # Define the bounding box for the area of interest
    crs=crs,   # Set the coordinate reference system
    bands=["tuw_likelihood","tuw_flood_extent"],   # Specify the bands to load
    resolution=resolution,   # Set the resolution of the data
    dtype='uint8',   # Define the data type
    chunks={"x": 1000, "y": 1000, "time": -1},  # Set the chunk size for Dask
)

Pre-processing

GFM_dc["tuw_flood_extent"] = GFM_dc.tuw_flood_extent.where(GFM_dc.tuw_flood_extent!=255).compute()

Plot

band_name = "tuw_flood_extent"
print(f"Visualizing band: {band_name}")

GFM_dc[band_name].hvplot.image(x="x", y="y", title=band_name, cmap=["#0000FF19", "darkred"])

Visualizing band: tuw_flood_extent